Part of the AI Regulation News hub.
Singapore's digital ministry has told Parliament that the age assurance obligations reaching AI chatbots sit on designated app stores rather than on chatbot developers, and that it is studying further safeguards for chatbots reached directly through websites
The question had two limbs. The answer engages one of them, and the phrases minor-safe defaults and crisis escalation do not appear anywhere in it.
Bottom line: This is a written answer to a parliamentary question, not an instrument. It binds nobody and changes no duty. It sets out where Singapore's existing chatbot safeguards sit today and says the Ministry is studying further ones.
Who this affects: Online safety and trust compliance officers at AI companion and mental health chatbot developers serving Singapore, policy teams at app stores designated under the Code of Practice, and child safety policy leads at digital health providers.
Issue date: 9 September 2026, the Parliament sitting of that date. Question for Written Answer No. 35, asked by Mr Yip Hon Weng.
What changed: Nothing in law. What is new is the Ministry saying on the record that some AI chatbots are reached directly through websites or other channels, and that it is studying further safeguards across modes of access.
Analysis: Both levers the answer names attach to app stores. The second limb of the question, whether common standards will apply across access channels, is not answered, and the answer gives no timeline, no instrument and no scope for whatever is being studied.
Primary sources: MDDI written answer on age assurance for AI companion and mental health chatbots
- Instrument (EN)
- Ministerial written answer on extending age assurance measures, minor-safe defaults and crisis escalation standards for high-risk AI companion and mental health chatbots across access channels
- Authority
- Ministry of Digital Development and Information, Singapore
- Jurisdiction
- Singapore
- Status
- Answered and published 9 September 2026
- Bindingness
- Non-binding. A written answer creates no obligation. The instruments it describes, the Broadcasting Act and the Code of Practice for Online Safety - App Distribution Services, carry their own force and were not opened for this piece
- Issue date / next deadline
- 9 September 2026. No consultation, no deadline and no commencement date is stated
- Document
- Parliament Sitting on 9 September 2026, Question for Written Answer, item 35
- Primary source
- https://www.mddi.gov.sg/newsroom/mddi-response-to-pq-on-extending-age-assurance-measures--minor-safe-defaults-and-crisis-escalation-standards-for-high-risk-ai-companion-and-mental-health-chatbots-across-access-channels/
What was asked, and what came back
The question, from Mr Yip Hon Weng, opened from a premise rather than a neutral enquiry. It put to the Minister that app store age assurance measures can be bypassed through web-based or account-free access at the point of download, then asked two things: why the Ministry has not required equivalent age assurance measures at the point of access to high-risk AI companion and mental health chatbots, and whether common age assurance, minor-safe defaults and crisis escalation standards will apply across access channels.
The answer is four short paragraphs. It opens by noting that AI chatbots can be accessed through different channels, including mobile apps distributed through app stores and services accessed directly through the Internet, and states that the app store is therefore an important gateway for access.
It closes by stating that some AI chatbots can also be accessed directly through websites or other channels, and that the Ministry is studying further safeguards to better protect younger users across different modes of access.
The two levers, and what they attach to
The Ministry states that the Broadcasting Act allows the Infocomm Media Development Authority to direct an app store to disable Singapore users' access to egregious content, and gives as an example apps that advocate or instruct on suicide or self-harm. The subject of that power is the app store, not the chatbot developer.
The second lever is the Code of Practice for Online Safety - App Distribution Services. The Ministry states that it requires designated app stores to implement age assurance measures and to minimise the risk of users, especially children, being exposed to harmful content, and that they are expected to review apps and app updates before release on their service to ensure compliance with the app store's content guidelines and standards.
Both of those attach at the same point. A chatbot that never appears in an app store never passes through either.
The gap the Ministry names itself
The closing paragraph is the reason this answer is worth a professional's attention. The Ministry does not dispute the questioner's premise. It states that some AI chatbots can also be accessed directly through websites or other channels, and that it is studying further safeguards.
That is a regulator acknowledging on the record that a category of access sits outside the levers it has just described. For a developer, the practical reading is that the perimeter as described today runs through distribution, and that the Ministry has said it is looking at that.
How much weight to put on it is a question of remove. An intention to study, which might later become a position, which might later become an instrument, is three steps from anything anyone has to comply with. We say all three because the distance is the point.
What the answer does not say
It does not answer the second limb. The question asked whether common age assurance, minor-safe defaults and crisis escalation standards will apply across access channels. The answer does not say whether they will. The phrases minor-safe defaults and crisis escalation do not appear in it at all, and neither does the phrase high-risk that the question used to frame the category.
It gives no timeline for the study, names no instrument that might carry the result, and describes no scope. It does not say whether a web-accessed chatbot falls inside or outside any existing code, and it does not say whether whatever is being studied would attach to the service, the access point or the distributor.
It also does not answer the first limb on its own terms. The question asked why the Ministry has not required equivalent measures at the point of access. The answer describes what is required at the app store gateway and says a study is under way. It does not give a reason.
None of that makes the answer evasive rather than ordinary. Written answers to parliamentary questions are short by convention. It does mean nobody should read a forthcoming rule into it.
What we did not verify
What we opened: the published written answer on the Ministry's newsroom, read in full, including the dateline of 9 September 2026, the sitting header, the question as tabled with its number and the name of the member who asked it, and all four paragraphs of the Answer.
What we did not open: the Broadcasting Act, the Code of Practice for Online Safety - App Distribution Services, the list of designated app stores, and the Singapore parliamentary Hansard record for the sitting. We describe both instruments only as the Ministry describes them in this answer, and we do not state what either actually requires.
What we refuse to claim: we do not say Singapore requires age assurance at the point of access to a chatbot, because the answer describes no such requirement. We do not say that web-accessed chatbots are unregulated in Singapore, because this answer addresses two instruments and we did not survey the rest of the statute book. We do not name a date, an instrument or a scope for the safeguards being studied, because the answer gives none. We do not say the Ministry accepted the premise that app store measures can be bypassed, because it neither accepted nor rejected it in terms.
One quotation note. The published answer uses typographic apostrophes, and en dashes both in the name of the Code of Practice for Online Safety - App Distribution Services and in the sentence about egregious content. Where we quote or name, we reproduce the words with ASCII punctuation, which is a house typesetting convention and not a change to any word, and we paraphrase rather than quote the sentence containing the dash.
Informational analysis for working professionals, not legal advice. Confirm how any rule applies to your situation with qualified counsel.
If you ship an AI companion or mental health chatbot into Singapore through an app store, the compliance surface described here is your distributor's, and it reaches your product through their review and their age assurance. If your users reach you through a browser instead, the Ministry has now said on the record that it is studying that route. Nothing has changed yet. What has changed is that the gap is documented in the Government's own words, which is a different starting position for the next consultation than an unacknowledged one.
Source File
Open the written answer and confirm three things: the two-limb question as tabled at item 35, the two instruments named in the second and third paragraphs of the Answer, and the final paragraph stating that the Ministry is studying further safeguards across different modes of access.
Some AI chatbots can also be accessed directly through websites or other channels. The Ministry of Digital Development and Information is studying further safeguards to better protect younger users across different modes of access. ยท MDDI written answer, Parliament Sitting on 9 September 2026, Question for Written Answer No. 35, final paragraph of the Answer
FAQ
Has Singapore introduced age assurance for AI chatbots accessed through websites?
No. This is a written answer to a parliamentary question and it introduces nothing. It describes age assurance obligations that the Ministry says attach to designated app stores under the Code of Practice for Online Safety - App Distribution Services, and states that the Ministry is studying further safeguards across different modes of access.
Does the answer say when new rules will arrive?
No. It gives no timeline, names no instrument and describes no scope. It says only that the Ministry is studying further safeguards to better protect younger users across different modes of access.
What did the Ministry say about suicide and self-harm content?
It states that the Broadcasting Act allows the Infocomm Media Development Authority to direct an app store to disable Singapore users' access to egregious content, and gives apps that advocate or instruct on suicide or self-harm as an example of that content. The power described operates on the app store.
Did the answer address crisis escalation standards?
It did not. The question asked whether common age assurance, minor-safe defaults and crisis escalation standards would apply across access channels. Neither phrase appears in the answer, and the answer does not say whether such standards will apply.
Related briefings
Sponsored Training
Practical AI training for regulated professionals, built around verification, documentation and a defensible process. See the courses.